Study and progress of anomaly target detection in hyperspectral imagery

Abstract

Hyperspectral image is a new kind of remote sensing images with the feature of"combining mapping and spectra into one",thus better expressing the subtle differences on the surface of the material through the continuous spectral curve. Hyperspectral images have a wide range of applications in such aspects as classification,unmixing and target detection. With the continuous development of hyperspectral remote sensing technology,anomaly target detection has become one of the most active direction of research because it doesn't need a priori information.Many anomaly target detection algorithms have been proposed. Based on data available both in China and abroad,this paper summarized the research situation and new progress in anomaly detection algorithms. The author first expounded the essence of hyperspectral anomaly target detection and used the basic theory and then analyzed and summed up some representative anomaly detection algorithms in such aspects as the ideas of algorithm,key technology,advantages and disadvantages. On such a basis,the author summarized and described the evaluation method of anomaly detection and discussed the future development trend of anomaly target detection algorithm,with the purpose of finding new breakthroughs in the study of the algorithm of hyperspectral anomaly target detection.